Skip to content

Latest commit

 

History

History
613 lines (425 loc) · 13.2 KB

faq.md

File metadata and controls

613 lines (425 loc) · 13.2 KB

Frequently Asked Questions

Common questions about Claude Code Tresor.

General Questions

What is Claude Code Tresor?

Claude Code Tresor is a comprehensive collection of professional-grade utilities for Claude Code CLI, including:

  • 8 Autonomous Skills - Automatic background helpers (code-reviewer, security-auditor, etc.)
  • 8 Specialized Agents - Expert sub-agents for deep analysis (@config-safety-reviewer, @test-engineer, etc.)
  • 4 Essential Commands - Workflow automation (/scaffold, /review, /test-gen, /docs-gen)
  • 20+ Prompt Templates - Production-ready prompts for common scenarios
  • Development Standards - Style guides, Git workflows, collaboration guidelines

Think of it as: A professional toolkit that augments Claude Code with specialized experts and automation.

Who created Claude Code Tresor?

Author: Alireza Rezvani License: MIT Repository: https://github.com/alirezarezvani/claude-code-tresor Contact: rezvani@gmail.com

Is Claude Code Tresor free?

Yes! Claude Code Tresor is open source under the MIT License:

  • ✅ Free for personal use
  • ✅ Free for commercial use
  • ✅ No attribution required (but appreciated!)
  • ✅ Modify and redistribute freely

Skills, Agents, and Commands

What's the difference between skills, agents, and commands?

Feature Skills Agents Commands
Invocation Automatic Manual (@agent) Manual (/command)
When to use Continuous monitoring Deep analysis Workflows
Tool access Limited (safe) Full access Orchestrates
Context Shared Separate Coordinates
Examples code-reviewer (skill) @config-safety-reviewer (agent) /review (command)

Simple explanation:

  • Skills = Automatic helpers (always watching)
  • Agents = Expert consultants (call when needed)
  • Commands = Workflow automation (multi-step processes)

When should I use a skill vs. an agent vs. a command?

Use Skills when:

  • ✅ You want automatic, continuous monitoring
  • ✅ You need real-time suggestions while coding
  • ✅ You want lightweight, non-blocking feedback

Example: Let code-reviewer skill monitor your code as you work.

Use Agents when:

  • ✅ You need deep, comprehensive analysis
  • ✅ You have a specific complex question
  • ✅ You need expert guidance on architecture/design
  • ✅ You're debugging a complex issue

Example: @systems-architect design a scalable microservices architecture

Use Commands when:

  • ✅ You need to automate a multi-step workflow
  • ✅ You want to coordinate multiple agents
  • ✅ You're running a standardized process (scaffolding, review, testing)

Example: /review --scope all --checks security,performance

How do skills work?

Skills are conversation-based autonomous helpers that activate automatically:

How they activate:

  1. You save a file with code changes
  2. Skill detects the change (based on file patterns)
  3. Skill analyzes the code
  4. Skill provides suggestions in conversation

Example:

You: [Saves UserProfile.tsx with missing error handling]

code-reviewer skill: "⚠️ Missing error handling in async function.
Suggestion: Add try-catch block to handle API failures."

Key points:

  • ✅ NO manual invocation needed
  • ✅ Works in background continuously
  • ✅ Provides suggestions without interrupting
  • ✅ Limited tool access for safety

How do I invoke an agent?

Use @agent-name syntax:

@config-safety-reviewer analyze src/api/users.controller.ts for security issues

Tips for better results:

  • Be specific about what you want
  • Provide context (production, requirements, constraints)
  • Specify focus areas

Example:

@systems-architect design user authentication system

Requirements:
- 10k concurrent users
- JWT tokens
- Refresh token mechanism
- Rate limiting

Constraints:
- Node.js/Express
- PostgreSQL
- 2-week timeline

How do I run a command?

Use /command-name syntax with options:

/review --scope staged --checks security,performance

Get help:

/review --help

Common commands:

# Scaffold components
/scaffold react-component UserProfile --typescript --tests

# Code review
/review --scope all --checks all

# Generate tests
/test-gen --file src/utils/helpers.ts --coverage 90

# Generate documentation
/docs-gen api --format openapi

Installation & Setup

How do I install Claude Code Tresor?

Full installation (recommended):

git clone https://github.com/alirezarezvani/claude-code-tresor.git
cd claude-code-tresor
./scripts/install.sh

Selective installation:

./scripts/install.sh --skills    # 8 skills only
./scripts/install.sh --agents    # 8 agents only
./scripts/install.sh --commands  # 4 commands only

Complete installation guide →

What gets installed where?

  • Skills: ~/.claude/skills/
  • Agents: ~/.claude/agents/
  • Commands: ~/.claude/commands/
  • Prompts: ~/.claude/prompts/
  • Standards: ~/.claude/standards/

Repository stays separate at ~/claude-code-tresor/ (or wherever you cloned it).

Do I need to keep the repository after installation?

Yes, recommended for:

  • Updates (git pull && ./scripts/update.sh)
  • Reference documentation
  • Examples and templates

But not required - installed utilities work independently.

How do I update Claude Code Tresor?

cd claude-code-tresor
git pull origin main
./scripts/update.sh

Preserves your customizations while updating utilities.

Can I uninstall Claude Code Tresor?

Yes:

# Remove all components
rm -rf ~/.claude/skills/code-reviewer
rm -rf ~/.claude/skills/test-generator
# ... (remove all skills)

rm -rf ~/.claude/agents/code-reviewer
# ... (remove all agents)

rm -rf ~/.claude/commands/scaffold
# ... (remove all commands)

# Remove repository
rm -rf ~/claude-code-tresor

Complete uninstall instructions →

Configuration & Customization

Can I customize skills/agents/commands?

Yes! Edit configuration files:

Skills:

vi ~/.claude/skills/code-reviewer/SKILL.md
# Edit YAML frontmatter

Agents:

vi ~/.claude/agents/code-reviewer/SKILL.md
# Edit YAML frontmatter

Commands:

vi ~/.claude/commands/review/command.json
# Edit JSON configuration

Complete configuration guide →

Can I create my own skills/agents/commands?

Absolutely! Follow our guides:

Can I disable specific skills?

Yes, set enabled: false in skill configuration:

---
name: "readme-updater"
enabled: false  # Skill disabled
---

Restart Claude Code for changes to take effect.

How do I configure file patterns for skills?

Edit skill's YAML frontmatter:

---
name: "code-reviewer"
file_patterns:
  - "src/**/*.ts"      # Only TypeScript in src/
  - "src/**/*.tsx"
exclude_patterns:
  - "*.test.*"         # Exclude tests
  - "*.stories.*"      # Exclude Storybook
  - "node_modules/**"  # Exclude dependencies
---

Troubleshooting

Skills not activating - why?

Common causes:

  1. Skill disabled:
---
enabled: false  # Change to true
---
  1. File patterns don't match:
---
file_patterns:
  - "*.ts"  # Does this match your files?
---
  1. File excluded:
---
exclude_patterns:
  - "node_modules/**"  # Is your file here?
---
  1. Need to restart: Restart Claude Code CLI

Complete troubleshooting →

Agent not found - what's wrong?

Solutions:

# 1. Check agent exists
ls ~/.claude/agents/code-reviewer/

# 2. Verify configuration
cat ~/.claude/agents/code-reviewer/SKILL.md

# 3. Check enabled field
---
enabled: true  # Must be true
---

# 4. Restart Claude Code

Complete troubleshooting →

Command not working - how to fix?

Solutions:

# 1. Check command exists
ls ~/.claude/commands/review/

# 2. Verify configuration
cat ~/.claude/commands/review/command.json

# 3. Try full path
/workflow/review  # Instead of /review

# 4. Check syntax
/review --help  # Show correct usage

# 5. Restart Claude Code

Complete troubleshooting →

Too many skill suggestions - how to reduce?

Option 1: Lower priority:

---
name: "code-reviewer"
priority: "low"  # Changed from "high"
---

Option 2: Disable temporarily:

---
name: "readme-updater"
enabled: false
---

Option 3: Narrow scope:

---
name: "test-generator"
file_patterns:
  - "src/**/*.ts"  # Only src/ directory
exclude_patterns:
  - "*.test.*"     # Exclude existing tests
---

Migration

How do I migrate from v1.0 to v2.0?

Major changes:

  • NEW: 8 autonomous skills
  • CHANGED: Agent syntax (rr- → @)
  • CHANGED: Directory structure

Migration steps:

cd claude-code-tresor
git pull origin main
./scripts/migrate-v1-to-v2.sh

Complete migration guide →

What's the new agent syntax?

v1.0 (old):

rr-code-reviewer analyze this

v2.0 (new):

@config-safety-reviewer analyze this

Performance & Limits

Do skills slow down Claude Code?

No - skills are lightweight:

  • Limited tool access (Read, Write, Edit, Grep, Glob only)
  • Efficient conversation-based activation
  • Run in background without blocking

If experiencing slowness:

  • Reduce active skills (disable non-essential)
  • Narrow file patterns (monitor fewer files)
  • Use Sonnet model instead of Opus for skills

How many skills/agents can I have?

No hard limit, but practical recommendations:

  • Skills: 5-10 active skills (more = more suggestions)
  • Agents: Unlimited (invoked manually as needed)
  • Commands: Unlimited

Balance: More skills = more feedback but potentially overwhelming.

Can I run multiple commands simultaneously?

Generally no - commands orchestrate workflows sequentially.

Exception: If commands are independent (different scopes).

Recommendation: Run commands one at a time for best results.

Contributing & Community

How can I contribute?

Ways to contribute:

  1. Report bugs - GitHub Issues →
  2. Suggest features - GitHub Discussions →
  3. Create skills/agents/commands - Contributing Guide →
  4. Improve documentation - Submit PRs for documentation
  5. Share prompts/templates - Contribute useful patterns
  6. Review PRs - Help review community contributions

Complete contributing guide →

Where can I get help?

Resources:

Is there a community?

Yes!

Advanced Topics

Can skills invoke agents?

Yes, through coordination configuration:

---
name: "code-reviewer"
coordination:
  invoke_agents:
    - agent: "@security-auditor"
      when: "security_issue_detected"
---

Future feature - currently skills suggest agent invocation to user.

Can agents invoke other agents?

Yes, using the Task tool:

@systems-architect design system
[Invokes @security-auditor for security review]
[Invokes @performance-tuner for optimization]

Can I use Claude Code Tresor with other Claude Code extensions?

Yes! Claude Code Tresor is designed to work alongside:

  • Other skills
  • Other agents
  • Other commands
  • MCP servers

Potential conflicts:

  • Skills with similar names (ensure unique names)
  • Commands with same name (use unique names)

Does Claude Code Tresor work offline?

Partially:

  • ✅ Configuration and documentation accessible offline
  • ✅ Templates and standards available offline
  • ❌ Skills/agents/commands require Claude AI (online)

Still Have Questions?

Can't find your answer?

  1. Search Documentation → - Browse complete docs
  2. Ask Community → - Post question
  3. Report Issue → - If you found a bug

Last Updated: November 7, 2025 | Version: 2.0.0